JPS62501497A - Single crystal silicon ingot drawn continuously - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention]

Continuously drawn single crystal silicon ingot Field of invention This invention relates to the silico part of a semiconductor semiconductor node, and is particularly applicable to the semiconductor node. :Okizuki Taketa 1.1 tone and 4 people in a monocrystalline doped silicon implant! ·of Person who goes to 9113 &: l! 11. . R”-Breath IW-BuV Semiconductor group 1 node silicon: 6-circuit 2-1' conductor, combination The foundation of the two industries is booming.゛1'-Conductor Grade Si-1 has five (characterized) high purity requirements. 7: It3, Jun 1 true level is J3 in other technical fields of chemistry and metallurgy) They are not sought after and cannot reach the bottom. ,,Semiconductor guide I- Another feature of silicone: 1. ;1,,,! iS! b) Purpose (here) , with a slight (J, but accurate (n known or]n1-") 3f) due to the necessity that silicon contains impurities. This is Siri “] The product is called doped silicon. Damn it, typical 4 doped into 1 silicon.

[Elements include boron, phosphorus, arsenic, and antimony. Boron is a typical electron collector, and phosphorus is a typical electron donor. For example, 1! ! Silicon doped with boron or phosphorus in certain regions or the entire silicon body, or silicon doped with boron or phosphorus in different regions of the silicon body, becomes a valve for electrons and is generally referred to as a semiconductor. It can be widely used in switches, amplifiers, bus J3 and other electronic functional components of suitable electronic i1 111 circuits. The semiconductor industry is highly developed today, with numerous semiconductor devices available directly on the market or incorporated into computers, radios, televisions, controllers, and a nearly countless variety of other electronic devices. The basis of this semiconductor industry is a relatively small, but still more The hand-maintained materials supply industry includes silicon and silicon compounds, compounds for doping silicon (called dopants), and doping. The company supplies silicon in its final or partially produced form, as well as other chemicals and various silicon-based components. One form of silicon widely used in the semiconductor industry is semiconductor gray. This is a single-crystal silicon ingot. Semiconductor grade, 111 crystalline silicon ingots are produced by a very well known and widely used classical method of growing single crystals known as the 1:1 xyls-4-co method. This method This is often called the C2 method. Single-crystal metal ingots are produced by growing a molten metal by bringing small single crystals into contact with each other according to the Czochralski method, and then slowly pulling the single crystal out of the molten metal while rotating the single crystal. It grows by . The single crystal is held at a temperature below the melting point of the crystal. The molten material near the crystal, which is in direct contact with the crystal, has an IQ of only a few atoms and is not a single crystal. They are piled up on crystal seeds and grow like seeds. The atoms in the melt have the same crystal structure as the seed. The crystals are piled up in the structure, and therefore larger single crystals are formed. This /process continues to occur layer upon layer of single crystal until a large ingot is formed. These ingots are very sensitive They are typically over 100 ko and can be several inches in diameter and several feet long. If the insert J'3 and the melt are When the material is a high purity metal such as semiconductor silicon, the crystal structure is essentially a perfect single crystal of 7:5 purity semiconductor grade silicon. Many improved methods J3 and variants of single crystal growth are known, see, for example, US Pat. However, the Czochralski method is not without its limitations, and if producing completely ultra-pure silicon is one of its two technological pinnacles, this method has serious problems. have. The Czochralski method is generally performed in a batch process. be exposed. A predetermined amount of silicon from a suitable source is melted in the crucible and the seed crystals are withdrawn until the melting crucible is empty. In the unlikely event that the silicon raw material is completely pure and If no impurities are introduced, it is theoretically possible to produce a perfect crystal of completely uniform purity, t≧A. However, this is not possible and the presence of even small impurities leads to inhomogeneities in the purity of the ingot. When the ingot is withdrawn from the melt, “bar icing” (dispensing) °′ A phenomenon known as oxidation occurs, where impurities either migrate preferentially into the crystal and grow, or preferentially remain in the melt. t'2'H is more common. Typically, the effect of bar eye staining tends to collect impurities in the melt. Therefore, as the crystal grows, the concentration of impurities in the melt increases. This distribution effect causes a relatively constant amount of impurities present in the melt to bind together. The level of impurities in the melt decreases during crystal growth, resulting in partitioning into the crystal. As the number of crystals increases, the level of impurities in the crystal also increases. Therefore C1 knot The level of impurities in a crystal increases as the crystal grows (the opposite occurs if impurities are preferentially distributed into the crystal). In devices that change the Purity gradients become a practical problem. This problem is often overcome by limiting the crystal size to a size such that impurity gradients along the crystal do not significantly affect the electronic properties. Already present throughout the crystal, whose electrical properties differ from those of ultra-high purity silicon? A crystalline ingot of silicon that is uniform at all locations along the length of the ingot, even though the suspension of the doban 1- is known or huntroled. l! It is desired to fj 9. As is clear from what has already been said, Dzikokralski's batch 1 is a homogeneously doped single crystal ingot. It is not very suitable for manufacturing commercially available silicon. Main Features of the Invention 1 1. According to Dzokralski-Brohiss, a continuous or nearly continuous variant, such that the ingot has a uniform level of dopant along the length of the ingot. A method for growing a lit crystal silicon ingot. Various methods are known for producing semiconductor grade silicon. Early methods of producing high-purity silicon included silane or silicon atom. Methods include pyrolyzing lachloride on a hot filament of tungsten, such as tantalum. Another such method is the pyrolysis of contetrachloride, named ``h method'' to obtain silicone. Generating recon. Silicon strongly adheres to the quartz tube and is recovered by breaking the quartz tube during cooling or during cooling. A variation of the process described above is described in U.S. Pat. It is In the quartz wall, the amount of silicon deposited on the quartz wall does not exceed 2 mils, and is broken down and peeled off from the wall during production. For the formation of silicon on nucleated silicon particles, see US Pat. No. 3,012,862, December 12, 1961, Partland et al. and US Pat. No. 3,012,861, 12F112, 1961, Ring et al. The generation of seed particles is discussed in US Pat. No. 4,207,360, June 10, 1980, by Badovani. For a variation of the pyrolysis process for silicon production, see U.S. Pat. No. 4,117,094, July 26, 1978, Broker Jr. It is listed in wet. Garagaglia et al., U.S. Pat. The manufacture of silicon for the main body is described. However, rather than a melt, it is a silicon melt. The surface of the silicon ribbon pulled out from The method of Galagaglia et al. This is a variation of the Russky method. How to produce ultra-pure semiconductor-grade silicon from tribromodylan No. 4,084,024, Simicef C. Schumer H-119 April 11, 1978, and US Pat. Thus, although prior art techniques are known for producing large ingots of silicon semiconductor material, whether by the Dzoklars method of growing single crystals from a melt or by the method of growing single crystals from a melt, heated filament 1- or vapor phase growth on ribbons of silicon, performed in pups, these processes are affected by the boundary conditions associated with batch processing and by impurity partitioning. There is a II+ limit. The present invention overcomes these prior limitations and inadequacies and provides continuous or near-continuous production of ultra-high purity semiconductor grade silicon ingots. This allows the production of doped silicon ingots with a high density and a certain doping level. Attempts have been made to produce single crystal silicon by a continuous Dicoclarus 4-knee process, but to the best of our knowledge no such process has been found to be fully reliable and completely satisfactory. One of the major problems with continuous crystal drawing methods of the type described above is that high quality, high purity single crystal growth is precisely dependent on maintaining a stable m and ripe balance in the melt. It's about being dependent. Even slight thermal disturbances can disrupt crystal growth and result in re-melting, usually as waste fruit. It produces a polycrystalline ingot that must be melted. This is obviously a very uneconomical and undesirable event. Prior art silicon feedstocks are typically made from large or small silicone bases or necks, which often vary widely in size, shape, and surface area. death Therefore, the handling of these materials is cumbersome and cannot be fed to the crucible by precise 81m means; the provision or attempt to provide a continuous feed system may affect the thermal balance J5 and stability of the Cz furnace. This causes incomplete crystal formation. There is a risk of long life! L decrease or increase. If silicon is crushed into fine powder, What if the surface area increases and the book simply comes from the grinding operation? 1 f, T impurities are added. Due to the large irregular surface area of such -F acids, it is possible to powder silicon without introducing much oxygen and other adsorbed and absorbed impurities into the melt. It is practically impossible to supply it in crushed or finely ground form. There is always a risk that the quality of the product will be compromised and you will end up with a large number of very expensive unwanted ingots. A single ingot is divided into thousands of wafers, and each wafer can contain tens to hundreds of semiconductor devices, or thousands of semiconductor devices formed therein. Considering that it is possible to create an integrated circuit with conductor devices, it is possible to create a complete system. Reliable method to produce single crystal ingots of recon! ! D gender is correctly recognized Should be. Important features of this invention (i.e. continuous production of silicon ingots) Eliminating most or all of the problems of conventional technology related to length, it is possible to economically produce a perfect monocrystalline silicon ingot of consistently high quality and high purity in an essentially continuous manner. This invention is characterized by <a) spherical particles with a low surface-to-volume ratio, monodisperse without individual agglomeration, and small diameter particles. About 1/2 to about 211,111 silicon particles are heated to 1JTi, (1) to form a silicon melt, and (C) the monodispersed silicon particles of step (a) are mixed into one culm (b). ) is continuously supplied with C, and then (d): a silicon single crystal is continuously drawn out from the silicon melt of (h) by about II, and J, - ) is added to the silicon single crystal over the entire length of the ingot. Ultra-n pure silicone with an evenly balanced composition Semiconductor grade silicon can be connected by forming an in-bore of silicon. Continuously manufactured. ? 'J method is described. This invention provides (a) forming a matrix of molten silicon; (1)) adding a body to the melt; Ultra-pure, spherical, non-individually aggregated monomers with a low surface to product ratio. (C) Dispersed silicon particles with a low surface-to-volume ratio, spherical, and not individually agglomerated; (d) producing particles of silicon containing a predetermined amount of dopant and having a diameter of about 1/2 to about 21, and (d) incorporating the particles of step (C) into the melt; Then, single crystal silicon having a constant φ of dopant is drawn out from the silicon melt containing <e> dopant, thereby forming an ingot. A constant level along its length is achieved by depositing an ingot of semiconductor grade silicon with a uniform degree of dopant along its length. The best method for continuously producing silicon ingots with high doping. is also advantageous. The invention also provides a doped silicon with a constant composition along its length. can be described as a method for the continuous production of ingots, i.e. two streams of spherical, low surface to volume ratio, monodispersed, non-agglomerated, silicon particles are fed into the silicon melt. one of the streams being of higher purity than the other stream, the other stream comprising dopant-containing silicon, and successively drawing single crystals from the melt. This invention involves a very important and unexpected development of the unique morphology of silicon, and a very impressive and unobvious use of such silicon in a novel way for the production of continuous perfect single crystal silicon ingots. It includes. ”! Bow 11 The Schuma cherSIIlcon process (5SP) described and claimed herein involves the unique shape of silicon. A new concept with new results that were previously unknown? It includes the recognition that it is possible to do something. Semiconductor grade silicon is spherical and has a small volume It can be made from particles of silicon that have a low proportion of surface area, are monodisperse without individual agglomeration, have an average diameter of about 1+i+i, and have a maximum particle size distribution of about 1/2 to about 21011. The product is such that each sphere is just about perfect and each sphere is identical. It is a free-flowing, "shot-like" product with approximately the same size as the particles. child The semiconductor grade used here refers to the semiconductor grade used in the semiconductor device manufacturing industry. For example, 0.1ppb boron, 0.31)lb phosphorus, etc. Such feedstocks are formed by pyrolysis or hydrogen reduction of bromosilane at 600°C to 1000°C under near atmospheric pressure in a "fluidized bed" reactor, with product diameters of 1/2 a. 2111a. Fluidized bed Raw material particles suitable for reactors can be made by grinding or abrasion of larger particles. However, from the point of view of their purity and avoidance from contamination, small base particles The morphology of the offspring is unimportant. The sphericity of the product of step 1 of the Schumarker silicone process develops during the deposition on the powder particles of these substrates fed to the fluidized bed reactor of step 1. Of course, the bromosilane Deposits occur as a result of pyrolysis or hydrogen reduction of the compound, producing silicon and various by-products depending on the exact bromosilane and the decomposition reduction method chosen.Sintering can be avoided in this process. This is important. In the bromosilane system, The escape of sintering is a low molecular weight that minimizes thermal decomposition! I decided to change the temperature of the hot water. can be achieved as a result of hydrogen reduction, e.g. This is facilitated by the dilution of hydrogen. T, thermal decomposition 4Sf I-+[3rs →Si +38+ Br4 +2112T214 z reduction S i HBr s+t12 →SI +311B'Here, T2 is higher than T. Sintering occurs as a result of surface diffusion under the driving force of surface curvature, producing a particle distribution that favors the growth of large particles with fewer small particles. It reduces the surface to volume ratio and binds the particles that are in contact with each other, further reducing the surface/volume ratio of the system.The product of Step 1 of the Schumarker Silicon Process (SSP) is monodisperse and approximately There are particles ranging from 1/211ffl to 211111. Particles are present or occur to a considerable extent. Here, Ti+p is the melting point of the sintered rod. Indicate the point ν′. In the case of silicon, since Tmp=1420° C., 1693′Kx 0.6−=Tsinterir+Q. However, at a temperature of approximately 750°C, sintering begins. Ru. However, in reality, since the reactor in Step 1 is a fluidized bed reactor, the contact time between one particle is λ1) and 1111, and for this reason a3 and the dispersibility of the product, there is some Sintering is not a problem up to moderate temperatures, i.e., about 1000 DEG C.-.1050 DEG C. In step 2, the dope and acceptor are added to the continuous Cz melt at Δ, and the doped silicon is made up to ¥J. The block [J model] used in this process Process 21 is similar to step 1, except that the working fluid of the dilan is not as pure as possible, but is taken from the toner or the part where the acceptor is concentrated. The prepared working fluids J5 and [3Br3 or PBrI are well-doped with boron or phosphorous and contain very little metal (441-grade) polycrystalline silicon for this purpose. Step 3 of a S S P , which is produced in a fluidized bed reactor and added to the feed stream, is a constant process by a Cz process that subtracts the seeds from the melt. Single bond of component It is a continuous or IF continuous production of crystalline silicon. The technology currently in use is the Siemens Process ``!1'' Straw Bladder: 11f, a process that does not use silane chemical reactions to achieve a specific IR (from 5 to 10 to 60 or 100 or more). Yes, it's 1-4! A semiconductor grade batch melt of ka) is produced and therefore the CZ feedstock cannot be produced continuously from steps 183 and 2, resulting in impurity partitioning across the solid-liquid layer boundary. . This means that impurities, including donors and acceptors, exist at different equilibrium degrees in a liquid in equilibrium with its solid phase. By being present. Therefore, molten FJ! A small batch of A-converted silicon goes through the CZ process. Since the impurity concentration changes with time in the melt, it also changes with location in the solid solidified from the melt. In SSP, this distribution is avoided for two reasons. In other words, (1) the batch process is not equal to 1, so the value of the parameter boundary is (2) the composition of the melt remains constant, and therefore the composition of the solid also remains constant, so it is independent of position in the solid (at least in the axial direction of the pull); about). The constancy of the volume of the melt is due to the addition of the SSP raw material, i.e. the product of step 1, and the solidification. Therefore, precise offset and deformation by the plate removed per unit time from the melt. It will be done. The key features of this invention are a controlled number of silicon particles that are spherical, have a low surface to volume ratio, are monodisperse rather than individually aggregated, and have an average diameter of about 11. It is a process in which silicon is added to the melt of silicon at a constant rate, typically sphere by sphere, essentially precisely replenishing the melt as the single crystal grows, but depending on the amount or thermal balance and the crucible. or sipping that disrupts or disturbs the stability of the melt. Solid layer n (well, the ratio of surface to volume is possible) The silicon particles are introduced in the smallest possible size and are characterized as spheres of silicon particles, individually non-agglomerated, monodisperse, and having a nearly uniform diameter of approximately Balancing, i.e. melting the added silicon and single bonding The effect on the heat required to compensate for crystal and environmental losses is a function of the rate of introduction of the individual particles. The rate at which one or more individual particles of essentially the same heat capacity are introduced at a time is regulated and held constant, so that the heat absorbed by the system also decreases. There are no disturbances or pass spikes in the heat required by the heat source. It is difficult to overestimate the importance of this aspect of the invention since more perfect single crystal silicon can be grown continuously and reliably. The molten mixture was doped into the SSP additive using the product of step 2 as raw r1. The desired composition of the produced ingot is kept essentially constant. Accordingly, the Schumarker silicon process comprises the following steps to produce superior quality wafers for the manufacture of semiconductor devices, including integrated circuits and silicon photovoltaic cells. 1. By fluidized bed pyrolysis of bromosilane, especially tribromosilane (St Her, ) at suitable Ul& (600°C to 1050°C), it is possible to obtain dense, monodisperse, spherical and large diameter (0.5 to 211111) Prepare beads of semiconductor grade polycrystalline silicon. 2. Prepare doped polycrystalline silicon balls of approximately the same size in the same manner as above except adding BI3r3 or PHI to the fluidized bed reactor. 3. Silicon and dope removed from the melt by solidification of single crystal silicon Add the products from steps 1 and 2 to the CZ melt in exactly the same amount as the punt. It is noted here that the large ball (approximately 110 ffi in diameter), together with its density, has a sufficiently low surface to volume ratio to eliminate contamination of the melt with excess oxygen. This is a known problem with older, smaller diameter and lower specific gravity fluidized bed products, which are used as feedstock for continuous or semi-continuous C7 of semiconductor grade silicon. This is something that is being attempted to be used in this way. In addition, the monodisperse quality of the products from steps 1 and 2 allows precise control of the thermal load placed on the system as the crystals are drawn and the beads are added to the melt. . The advantage of this process is that it gives wafers of improved quality, which This must be due to the fact that the ingot into which the wafers are divided is as uniform as possible. stomach. This uniformity is enhanced by the continuous or semi-continuous withdrawal of CZ crystals from the melt with a constant gradient and a constant component gradient, resulting in a slight increase in the lj component. misfortune happens. A result of about 300 feet 1- is reached before the increase in trace components becomes too large. You should be able to pull out the crystal. In fact, the melt added to the melt actually melts less than 3 tons before contacting the growing solid-liquid interface. Therefore, the additive (weir) matl- is hidden in other arrangements, and before it is carried by convection to the region of the growing solid-liquid interface, the additive (weir) matl- is a i-61;, enough time is provided be done. Seki's arrangement is the only characteristic of the SSP melting pot that is distinguished by the usual CZ rules. It is a sign. Therefore, compared to the C7 crystalline extraction equipment, new equipment is not required. Therefore, among the two patents on crystal growth technology, there is a general explanation. iQbi as specified and already mentioned (!i is used only with slight modification) It is something that Crystals 1.1 are drawn from shallow or deep melts. Divine forms such as heating, induction, resistance, RF, and microphones can be used. Electromagnetic fields can be used to control wall contact and melt convection paths. All of these are semiconductor grade! crystal The equipment in which SSP operates, known as Linney's CZ drawing technology, differs from standard crystal drawing furnaces only in its arrangement for handling the ingot as it is drawn from the melt. ing. Standard towers only handle ingots of limited length. In SSP, the pressure from the melt is removed by lateral supports which are also installed to support at least a partial load. The array is EQ-shaped so that the crystals are kept in a straight line within the boundary angle. The produced ingots are significantly longer than negative-type ingots and are much less variable with the location of flIa and abundant solute species than conventional CZ ingots. It has become. The ingots are then sliced, stacked and wafered in the conventional manner. These wafers are free of doping, defects or impurity concentrations. There is no significant change in iJ from one wafer to the next, and from one end of the ink to the other, clearly representing a significant improvement from a technical standpoint.衷/7i(1i111) Ultra-high purity monodisperse spherical silicon particles are described in U.S. Pat. No. 4,084,942. It was built according to the process described. The particles are dense, approximately uniform in size from 1/2 to 2B, primarily about 11 in diameter, and flow free of fines and debris. A dynamic product was formed. Using a method similar to that described in U.S. Pat. No. 4,084,942N, but dosing Punt, such as boron tribromide or phosphorus tribromide, is controlled by 111. Using tribromosilane as a raw material at a given concentration, a product with the same physical appearance, size and characteristics, but containing a known uniform degree of dopant, was prepared. The concentration of the dopant can be scaled to a given T and effective level; Typically in the range of 0.001 to i ppm, this product is used as a feedstock in silicon melts to provide low levels of dopants. Relatively high dopant concentrations are often used. For eel, there are various ratios of raw materials. The concentration of dopants in the silicon melt can be controlled by 65 because it can be read. Finally, the molten silicon matrix is finished and homogenized with the raw materials from the previous step, as well as the desired ratio of high-purity silicon and doped silicon. A melt of uniform composition is formed. The composition is controlled throughout all operations of the process by continuous feeding. The desired original VL ratio is maintained uniformly throughout the production. 5ppb (7) F-Pa The melt can be maintained by supplying equal h% of high purity silicon below o,'vppb and doped silicon with 1o ppb of dopants. held. Monocrystalline silicon (high IIT! degrees is also fine) is used in the classic Boo, Yokurals Follow the key method (contact the melt, then rotate at 6i, pull out for a long time, and A single-crystal ingot with a uniform composition of 5 ppb over its entire length was grown. let ? 11i crystals can be continuously drawn from the melt for very long periods of time. Although it is said that a crystal of this length cannot be handled in one minute due to the height of the equipment and handling considerations, a single crystal with a length of 300 feet or more can be completely It has been shown that it is possible. In summary, the method of the present invention involves forming a melt of silicon metal of two raw materials of + silicon such that one of the raw materials contains a predetermined level of doban. and continuously forming a single crystal ingot. Then, the doped silicon 11i crystal ingot 1 is continuously drawn from the silicon melt, and the ingot is injected!・Concentration of dopant along the length The raw material is continuously fed into the silicon melt, thereby controlling the dopane I- in the melt while the single crystal is drawn from it. This is to maintain a uniform concentration. As used herein, the term "continuously" or "continuously" refers to a process that repeats a sequence of steps periodically or without interruption. Thus, although sometimes referred to as semi-continuous, a continuous process involves the process of removing crystals from the melt, as well as the constant addition of silicon material during the pulling process of the raw material. The silicon material raw material is periodically introduced while drawing out the material. It also includes. Since all raw materials inherently contain some impurities or additives, the terms ultra-high purity silicon and doped silicon are used in their general technical sense. Within the scope of this invention In other words, the two feedstocks can each contain two concentrations of the same dopant, or two or more dopants with the same or different Q degrees.・The method of the second invention uses the first raw material, ultra-high axiality semiconductor grade silicon. A first raw material having a known ω is supplied to a melting body of silicon metal. 1' Supplying a second feedstock comprising ultra-pure semiconductor grade silicone doped with conductor dopants (and removing dopant from the melt while proceeding with the steps described above). Preferably, the method of the present invention is a spherical e, surface to volume ratio. A first silicone mixture characterized by J, -3 is formed into silicon particles with a diameter of low, individually agglomerated (not agglomerated, monodispersed, and approximately 1). Introduced into a molten potato, the ratio of the surface to the spherical rS volume is low, and the particles condense individually. doped silica L/(, 13, monodisperse, with a diameter of about 11!ffi) A second silicon mixture characterized by 32 particles is introduced into the melt, and while these steps are proceeding, the doped silicon is simultaneously removed from the melt. Contains one culm called a drawer made of semi-conductor grade ingot. Preferably, the Lt doped silicon particles contain boron, antimony, arsenic, or phosphorus. In a particularly preferred embodiment of the invention, the silicon particles and doped silicon particles have an average diameter of about 1,111,111 mm, and the particle source consists of essentially sintered particles, essentially having a diameter of 2 am or more. It is essentially free of particles and essentially fine particles with a 111 diameter of 1/2 mm or less. Lmatayu! This invention is widely and commonly used in the semiconductor industry. international search report

Claims (8)

[Claims] 1. (a) Two silicon raw materials in which one raw material contains a predetermined amount of dopant. (b) doped silicon from said silicon melt; of single crystal ingots are continuously drawn out, and said ingots are It is characterized by a uniform dopant concentration throughout the (c) continuously feeding said raw material to said silicon melt, thereby Maintaining a uniform concentration of dopant in the melt while drawing a single crystal from A method for continuously producing a single crystal ingot, comprising the steps of: 2. (a) A first raw material is supplied to a silicon metal melt, and the first raw material has a super high temperature. Featuring purity semiconductor grade silicon, (b) supplying a second raw material to the silicon melt, and supplying the second raw material to the semiconductor dope; Featuring ultra-pure semiconductor-grade silicon doped with a known amount of Punt, plus (c) While proceeding with steps (a) and (b), doped silica is removed from said melt. Steps (a), (b) and (c) proceed essentially sequentially. A method comprising the steps of: 3. (a) forming a melt of semiconductor grade silicon; (b) the melt is spherical, has a low surface-to-volume ratio, and is individually agglomerated; introducing monodispersed silicon particles having a diameter of about 1 mm, (c) the melt is spherical, has a low surface-to-volume ratio, and is individually agglomerated. doped silicon particles with a diameter of approximately 1 mm are introduced. , and (d) while proceeding essentially sequentially with steps (b) and (c). , essentially continuously producing semiconductor grade doped silicon from the melt. A method for growing single-crystal silicon, which includes a step of pulling out a nanogot. 4. Doped silicon particles are boron tribromide or phosphorus tribromide 4. A method according to claim 3, characterized in that the method comprises: 5. The silicon particles of step (b) and the doped silicon particles of step (c) are , having an average diameter of about 1 mm, the particle raw material is essentially a sintered particle, having an average diameter of about 1 mm. Does not contain particles with a diameter of 2 mm or more and essentially fine particles with a diameter of 1/2 mm or less. 4. A method according to claim 3, characterized in that: 6. (a) forming a melt of silicon; (b) forming a single crystal of silicon from said melt; Pull out the ingot and (c) a spherical, relative to the volume of the melt so as to be able to draw the silicon therefrom; It has a low proportion of surface area, is not individually aggregated, is monodisperse, and has a diameter of about 1/2 to Controls individual silicon particles, characterized as silicon particles up to approximately 2 mm. A method for growing single crystal silicon ingots with each step being introduced at a controlled rate. . 7. (a) forms a melt of silicon; (b) is spherical and has a surface to volume ratio; silica with low particle diameter, non-individual agglomerates, monodisperse, and essentially uniform diameter. Individual silicon particles, characterized as particles, are introduced into the melt in a controlled manner. Essentially continuous introduction at speed and further (c) the rate of introduction of silicon as particles as described above essentially makes it possible to halve from said melt; Single-crystal, with each step of drawing a single-crystal ingot of conductor-grade silicon. Continuous growth method for silicon ingots. 8. Furthermore, (d) it is spherical, has a low surface-to-volume ratio, and is not individually aggregated. dopant-containing silicon grains that are monodisperse and essentially uniform in diameter. controlled individual silicon particles different from the first said particle characterized as a child; Claim 6, characterized in that it comprises the step of introducing the melt into the melt at a rate of The method described in section.